Learning & Innovation Research Manager| Project Management Institute (PMI)Spain
Data quality and quantity is particularly important as we think about leveraging AI on projects. Considerations include the diversity and comprehensiveness of the data that is available to us.
Have you ever encountered unexpected challenges or pitfalls while using data in your projects? How did you navigate the situation and find a resolution? Saving Changes...
We collect financial data for business modeling for healthcare clients. We typically find challenges in data based on the consumers assumption and interpretation of their data. After we analyze the data, we have to adjust our predictions based on more accurate data. It's typically in the form of net % collection of revenue - what was billed vs. what was collected from payors. In almost all case, what was contractually agreed upon does not always show up in the financial data, so it's imperative this data is reviewed. Saving Changes...
Kenneth PettyIT Project Manager| Lincoln Financial GroupFort Wayne, In, United States
All Large Language Models are only as good as the quality of the data and quantity of that good data. The approach we have used is data cleansing and then feeding only verified quality data to train our models and not use internet or other public sources to avoid the issues with data ownership that these public models are running into now. The model needs to continue to be fed with data that has been curated for quality and breadth and as the data fed to the model increases the responses the model produces to the prompts improve. If we encounter bad responses we roll back to a prior model (saved with the data it was trained on) and start over with reviews of the data that was fed into the LLM that caused the problem. Automated prompt testing routines run nightly can help with issue identification. Saving Changes...
HAMEED ALYAJORIAudit and Risk General Manager , Project Manager| Yemen Customs AuthoritySn, Yemen
Yes, I’ve encountered unexpected data challenges in projects, particularly related to incomplete, biased, or inconsistent data. One case involved a risk analysis project where historical customs data was used to predict potential fraud cases. The dataset had missing fields, outdated classifications, and regional biases, leading to skewed AI model results.
How I Navigated the Challenge:
Data Cleaning & Imputation – I worked on filling gaps using statistical imputation and domain-specific rules, ensuring missing data didn't distort predictions.
Bias Identification & Correction – I analyzed the data distribution and applied reweighting techniques to balance underrepresented segments.
Cross-validation with Experts – I collaborated with customs officers and risk analysts to validate AI-driven insights against real-world trends.
Continuous Monitoring – Implemented an adaptive feedback loop, refining the dataset as new, more diverse data was integrated.
Ultimately, recognizing the limitations of the data early and iteratively refining the dataset helped improve AI-driven decision-making. Have you faced similar data challenges in your work?
The Data Landscape of GenAI for Project Management course was very eye opening! There are risk and limitations in managing data and this is amplified with new technologies. The inclusion of innovations sparks a response in general population the Gartner Hype Cycle is an illustration of the peaks and valleys of this phenomenon.
The best way to manage these risk is to mitigate using an iterative process. This is not limited to: data protection regulations respects user privacy, identifying biases in data outputs, and clarifying the data’s ownership and control. Saving Changes...
Data in particular available as guidelines in Statutory regulations / Standards change or gets updated time-to-time. It is imperative that one needs to keep abreast with the changes / amendments.
Use of data from the superseded standards/regulations will invite non-compliance & hence puts the project in to risk. Such instances occurred in few projects which had an implication on cost as well as schedule.
Leveraging the benefit of AIs on such instances can help overcome the pitfalls. Saving Changes...
It's important to collaborate with a data expert to define the parameters for data collection. Ensuring we are collecting data based on our question or goal to avoid pitfalls and challenges during a project. When a challenge does arise, it's important to reconnect with your data expert to enact a solution. Saving Changes...
In my previous project, the objective was to reduce obsolescence for database technologies across the globe aka many regions. The timeframe for communication with immense amount of data took two weeks cycle for response and plan of action. However the data kept growing as many assets were not listed in a particular region as API failed to retrieve information at the right time. Hence I took that region as a subset project with dedicated resources and fast tracked to bring the region at par with other regions. It was immensely challenging but at the end it was rewarding in terms of achievement and stakeholders appreciation.
Amazing work! Saving Changes...
Anonymous
We are using GenAl for our projects Saving Changes...
In my own case,during our recent SWOT analysis survey project,I identified inconsistencies and irregularities in the data collection process.Upon noticing unusually high volumes of data entries, I promptly collaborated with team members to investigate the anomaly. By monitoring input trends and conducting detailed analysis, we pinpointed the source of the discrepancies. I maintained open communication with stakeholders, providing regular updates on the issue and our mitigation efforts, which helped minimize the project’s overall impact. This experience underscored the importance of real-time monitoring of data inputs and usage to swiftly detect anomalous patterns or behaviours,a proactive approach critical to addressing risks before they escalate. Saving Changes...
Thomas MoyerGroup Media Director| H&L PartnersSan Francisco, Ca, United States
We use data for advertising purposes, feeding it in via LiveRamp into ad platforms. Our challenge is having quality data with high match rates in those ad platforms. Ultimately, we worked with the data team scrub that data to make it more valuable when we leverages it in the future. Saving Changes...
PMI Team Member
